AMERICAN MUSEUM Novitates

Transcription

1 AMERICAN MUSEUM Novitates PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y Number 2862, pp. 1-24, figs. 1-10, tables 1-6 October 31, 1986 Sundaic Rattus: Definitions of Rattus baluensis and Rattus korinchi GUY G. MUSSER' ABSTRACT Among the 14 genera and 41 species of murid rodents that are native to the peninsula and islands on the Sunda Shelf are five species ofrattus. Two of these, R. baluensis from Gunung Kinabalu in Sabah and R. korinchi from the mountains ofcentral Sumatra, were regarded as isolated mountain populations of one species. Restudy of specimens in museum collections indicates the Bornean population to be a species closely related to the widespread Sundaic R. tiomanicus, which is a relative ofr. rattus and other species native to continental Asia and India. Morphological characteristics of R. korinchi are unlike any other species ofrattus. These endemic Bornean and Sumatran Rattus are contrasted and defined. INTRODUCTION Peninsular Thailand and Malaysia south Thirty-three of the 40 occur there and nowhere else and their evolutionary histories of about latitude 10 N; the large islands of Sumatra, Borneo, Java, Bali, Palawan; and have probably been closely associated with numerous smaller islands are all part of the the Shelf; ten species are found on the peninsula and several large and small islands; great Sunda Shelf, an extensive 1,850,000 square kilometer extension of the Asian continent (Chasen, 1940; Tjia, 1980). In their one ofthe large Sunda islands, or the Palawan and 23 are endemic to either the peninsula, report on Malaysian murids, Musser and region. Seven co-occur on the Shelf and in Newcomb (1983) recorded 40 species of rats Indochina. The genera Lenothrix, Pithecheir, and mice, representing 14 genera, native to Sundamys, Kadarsanomys, and Palawanomys are endemic to the Shelf; the peninsula and islands of the Sunda Shelf. Maxomys, I Archbold Curator, Department of Mammalogy, American Museum of Natural History. Copyright K American Museum of Natural History 1986 ISSN / Price $2.75

2 2 AMERICAN MUSEUM NOVITATES NO Chiropodomys, and Haeromys are not entirely restricted to the Shelf but most of the species in each is found there; and Niviventer, Berylmys, Mus, and Rattus have few representatives on the peninsula and islands of the Shelf. Musser and Newcomb (1983) recognized only 4 of the more than 50 species of Rattus as native to the Sunda Shelf: R. tiomanicus, R. annandalei, R. hoogerwerfi, and R. baluensis. Rattus tiomanicus is widespread on the Shelf and exhibits appreciable insular variation in body size and fur color. Rattus annandalei occurs on the Malay Peninsula, Sumatra, and a few small islands. Rattus hoogerwerfi is known only from the Aceh region in northern Sumatra. Rattus baluensis was thought to consist of a population on Gunung Kinabalu in northern Borneo (R. R. baluensis) and a population in the mountains of Sumatra (R. R. korinchi). When Cameron Newcomb and I wrote our report, however, I was impressed by the differences in certain cranial and dental dimensions between samples from Gunung Kinabalu and Sumatra, all ofwhich we had called R. baluensis. Our assessment of that geographic variation was based upon my notes and measurements made several years earlier; we did not have the opportunity to reexamine specimens before the paper was finished. Shortly after our report was published, I looked again at the large series of baluensis available in museum collections and the few examples of korinchi in order to test the hypothesis that the Bornean baluensis is more closely related to another Bornean species and that korinchi is a derivative of a Sumatran species. At about the same time, I received a stimulating letter from the Earl of Cranbrook, an excellent biologist who has contributed much to our knowledge of Sundaic mammals. He mentioned the great distributional gap between the Bornean and Sumatran baluensis and wrote that "it seems to me far more likely that convergent forms, locally derived from tiomanicus ancestors, have evolved on high mountain peaks. Characters such as the long fur are likely to be adaptations to the cool climate." I appreciate the Earl of Cranbrook's independent insight into the problem. He inspired me to carefully study the Sundaic specimens and to also look again at character variation among other species of Rattus, a task that seems never to end. The results are redefinitions of Rattus baluensis and Rattus korinchi. The Bornean species shares many characters with the widespread R. tiomanicus and the relationship between these two reflects the picture suggested by the Earl of Cranbrook. Rattus korinchi, on the other hand, is not closely allied to R. baluensis, R. tiomanicus, or any other species ofrattus. It is geographically isolated in the mountain forests of Sumatra and appears to be morphologically unique within the present definition of the genus. ABBREVIATIONS AND PROCEDURES Specimens examined and referred to here are in collections of the American Museum of Natural History, New York (AMNH); the Philadelphia Academy of Natural Sciences, Philadelphia (ANSP); the British Museum (Natural History), London (BM); the Field Museum of Natural History, Chicago (FMNH); the Museum Zoologicum Bogoriense, Bogor (MZB); the Museum of Comparative Zoology at Harvard College, Cambridge (MCZ); the Rijksmuseum van Natuurlijke Historie, Leiden (RMNH); and the National Museum of Natural History, Smithsonian Institution, Washington, D.C. (USNM). Values for total length and length of tail are those recorded by collectors on skin labels. I subtracted length of tail from total length to obtain length of head and body. Values for length ofhind foot (including claw) were either taken from skin labels or from my measurements of dry skins. Cranial and dental measurements were taken with dial calipers graduated to tenths of millimeters. Their limits are defined in Musser and Newcomb (1983) with the exception that their "length ofpalatal bridge" is here called length of bony palate. The nomenclature I use for cusps and cusplets of upper and lower molars is presented in figure 1. See Musser and Newcomb (1983) for explanations of geographic names. Here I use sungai (stream or small river), gunung (mountain), and pulau (island).

3 1 986 MUSSER: SUNDAIC RATTUS 3 ACKNOWLEDGMENTS My reliance upon specimens makes me continually appreciate curators and supporting staff of the museums listed above who care for specimens and who have always generously allowed me to study and borrow material. I remain indebted to these responsible persons and grateful for their support. Mr. Peter Goldberg is responsible for the photographs ofcrania and teeth, Ms. Patricia Wynne drew figures 9 and 10, and Ms. Lauren Duffy produced the scanning electron micrographs in figure 8. I appreciate the fine efforts of these artists. The report has been strengthened by reviewer's evaluations, particularly those from Ms. Elizabeth Strasser, Dr. Michael D. Carleton, whose pertinent comments improved my prose without being really obnoxious; also, Dr. Marian Dagosto, Dr. M. Raymond Lee, Mr. Richard L. Green, and the Earl of Cranbrook. This report forms Results of the Archbold Expeditions Number 1 7. i bis (2 2 bis a-cen aln a-lab - d PIC Fg 1. hd- srcu-ed PC a-lab PCd pd mo 7 P~~~~C h-pc L --3 a-lab RATTUS KORINCHI AND RATTUS BALUENSIS Among the specimens of birds and mammals collected by Herbert C. Robinson and C. Boden Kloss during their expedition to Korinchi Peak in western Sumatra (now known as Gunung Kerinci, which is in Propinsi Jambi on modern Indonesian maps) were two rats obtained at Sungai Kring, on Gunung Kerinci, 7300 ft, in April 1914 (fig. 2). An adult and immature female formed the basis of Robinson and Kloss's (1916, p. 275) description ofepimys (Rattus) korinchi, which they briefly characterized as "Like E. baluensis (Thomas) with long, soft and spineless fur, beset on the upper surface with numerous longer piles; but with longer tail and paler underparts; nasals broader, but bullae much smaller; teeth considerably larger." Two years later, Robinson and Kloss (1918, p. 54) reported all of the mammals collected on the expedition to Gunung Kerinci, and listed korinchi as a subspecies ofrattus baluensis, writing that the rat "is the Sumatran representative ofr. baluensis." Robinson and 4.v,~~~~46pd- ~~md prtcndh,dd; ed, ~ cordng o Mlle's PC 191)shd-eanreferred to Fig. 1. Nomenclature of dental structures: a diagram of upper and lower molars in Lenothrix canus. Upper molars: cusps are numbered according to Miller's (1912) scheme and referred to in the text with the prefix t; pc, posterior cingulum. Lower molars: a-cen,, anterocentral cusp; a-lab, anterolabial cusp; a-ling, anterolingual cusp; pd, protoconid; hd, hypoconid; md,, metaconid; ed, entoconid; pc, posterior cingulum; alc, anterior labial cusplet; plc, posterior labial cusplet. Kloss knew Rattus baluensis by a few specimens collected from the slopes of Gunung Kinabalu in Sabah, where we now know the species to be common at altitudes between 7000 and 12,500 ft (Medway, 1977; Musser and Newcomb, 1983). In their account of Rattus baluensis korinchi, Robinson and Kloss provided short descriptions of fur coloration, skull conformation, teeth, and listed some measurements. During the following year, Robinson and Kloss (1919, p. 315) wrote about the mammals obtained from the Ophir District of western Sumatra (identified as Propinsi Sumatera Barat on modern maps) by E. Jacobson. Among them were two adults of what Robinson and Kloss identified as Rattus baluensis korinchi caught near the summit of Gunung Talamau, 9100 ft (known as Mount Ophir in the old literature and as Gunung Talakmau or Talaimu on modern maps), which is northwest of Gunung Kerinci (fig. 2). Both specimens were obtained on June 14, Robinson and Kloss noted that the "specimens agree closely with the type having

4 4 AMERICAN MUSEUM NOVITATES NO Fig. 2. Geographic distributions of three Sundaic species. Rattus baluensis: 1, Gunung Kinabalu, Sabah. Rattus korinchi: 2, Gunung Kerinci; 3, Gunung Talakmau. Rattus hoogerwerfi: 4, Gunung Leuser. the same long spineless pelage, with woolly underfur, ochraceous clay above, many ofthe longer hairs with broad black tips. Beneath greyish, the breast median line whiter, all the pelage grey at the base." Reference to Rattus baluensis korinchi appeared again in the literature with the publication of Robinson and Kloss's Addenda and Corrigenda to the mammals of Korinchi and Sumatra in 1923 where they provided a very good illustration of the skull of the holotype (p. 316, pl. III). From the early 1900s until now, korinchi has been treated as a subspecies of Rattus baluensis (Chasen, 1940; Ellerman, 1941; Musser and Newcomb, 1983), an association based upon the early allocation made by Robinson and Kloss and not by critical study of specimens. The four examples recorded in the reports by Robinson and Kloss are the only specimens ever assigned to korinchi. The other published record of korinchi proved to be a misidentification (see p. 352 in Musser and Newcomb, 1983). I have been able to locate only two of them: the holotype from Gunung Kerinci, which is in the British Museum (Nat. Hist.) where it is registered as ; and one of E. Jacobson's specimens (no. 351) from Gunung Talakmau, which is housed in the Rijksmuseum van Natuurlijke Historie and registered as I know ofno new material collected after Firsthand study of these two adults indicates to me that korinchi is a distinct species that is not closely related to Rattus baluensis. The first published reference to what would

5 1 986 MUSSER: SUNDAIC RATTUS 5 later be known as Rattus baluensis is by Thomas, who in 1889 recorded four specimens under the name Mus rattus that had been collected by John Whitehead during his expedition to Gunung Kinabalu in northern Borneo (fig. 2) in Two ofthe specimens came from 8000 ft and two from 3000 ft. Thomas (1889, p. 233) wrote that the "two specimens from an altitude of 8000 feet have their fur long and soft, while in those from 3000 it is short and harsh, so that it seems difficult to believe that both the forms can be referable to the same species." By 1894, when "A Preliminary Revision of the Bornean Species ofthe Genus Mus" appeared, Thomas (1894, p. 454) had examined specimens collected by A. Everett from 7000 to 10,000 ft on the slopes of Gunung Kinabalu in addition to those obtained by Whitehead at 8000 ft and was convinced that all examples represented a new species, Mus baluensis, which he characterized as being "Very similar to M. neglectus in size, proportions, and colour, but the fur quite long and soft, and the belly of a uniform dirty buff hue." This Rattus ofthe Kinabalu mountain forest has been recognized as a species by some workers (Sody, 1941; Ellerman, 1941; Misonne, 1969) but not others. Chasen (1940, p. 158), who included baluensis in his handlist of Malaysian mammals, wrote that it was "Almost certainly only a high level form of R. rattus," which in his view translated to a mountain subspecies of R. rattus. Allen and Coolidge (1940, p. 163) did list baluensis as a subspecies of R. rattus in their report on the mammals collected during the Asiatic Primate Expedition. While Ellerman (1949, p. 61) also placed baluensis with R. rattus, he noted that it "is an aberrant form, and its former race, korinchi is still more aberrant." Medway (1965, pp ), however, in his annotated checklist of the mammals of Borneo, reaffirmed the morphological and ecological integrity of Rattus baluensis, writing that the species "is a very distinct form, separated from other members of the R. rattus group by both morphology and habit. It must be considered a good species, and is in fact exceptional in that it is the only species ofthe group occurring in Borneo which is not in the slightest degree commensal with man." The distinctive nature ofrattus baluensis was also recognized by Musser and Newcomb (1983), and I continue to view it as a species different from R. rattus. My study of Rattus baluensis included 78 specimens (BM , , , (holotype), and ; FMNH 49267, , and ; MCZ ; USNM , , , , , , and ) from an altitudinal range of ,500 ft on the slopes ofgunung Kinabalu in northern Borneo (now Sabah), where it appears to be endemic. Rattus baluensis must be common if samples in museum collections are reliable indicators of its abundance in the wild. Allen and Coolidge (1940, p. 163), for example, reported that the collector of the series they studied noted R. baluensis to occur only "at higher altitudes; very plentiful, from 9 to 11,000 ft. This rat is quite tame, running over your face or eating your food if you sleep out on the ground." Information on tags attached to study skins indicates most specimens to have been caught in "moss" and "heath" forests. Lim and Heyneman (1968, p. 259) collected the rats in a zone between 7040 and 8900 ft, which they termed "mossy forest," and a higher zone between 9800 and 11,200 ft, which they referred to as "sub-summit dwarf forest and scrub." The entire altitudinal zone from which examples of R. baluensis have been obtained is montane forest, partially described by Jacobs (1961) and more thoroughly by Corner (1964), who also provided excellent photographs of montane forest on Gunung Kinabalu. I have no information about habitat or habits ofrattus korinchi except that the places where the rats were caught (7300 ft on Gunung Kerinci, 9100 ft on Gunung Talakmau) are in montane or moss forest (see Whitten et al., 1984; Whitmore, 1984). The two specimens of korinchi from Sumatra represent a species of Rattus with a moderately large body, a long tail relative to length of head and body, dense and soft fur, large molars, and small auditory bullae (table 1). Fur covering the upperparts of head and body is dark tawny, that on the underparts is grayish white with large white patches on

6 6 AMERICAN MUSEUM NOVITATES NO TABLE 1 Measurements (in Millimeters) of Adult Rattus baluensis from Sabah, Rattus korinchi from Northwestern Sumatra, and Rattus hoogerwerfi from Northern Sumatraa R. korinchi BM RMNH Measurement R. baluensis lb R. hoogerwerfi Length of head and body ± 8.4 ( ) ± 6.7 ( ) 20 Length of tail ± 8.7 ( ) ± 11.7( ) 20 Length of hind foot 32.4 ± 1.4 (30-35) ± 0.9 (36-39) 20 Greatest length of skull 40.8 ± 1.3 ( ) ± 0.8 ( ) 16 Zygomatic breadth 19.1 ± 0.6 ( ) ± 0.6 ( ) 18 Interorbital breadth 6.2 ± 0.2 ( ) ± 0.2 ( ) 20 Length of rostrum 12.9 ± 0.7 ( ) ± 0.5 ( ) 17 Breadth of rostrum 7.5 ± 0.3 ( ) ± 0.3 ( ) 18 Breadth of zygomatic plate ( ) ( ) 20 Depth of zygomatic notch 1.9 ± 0.2 ( ) ± 0.3 ( ) 20 Breadth of braincase 16.6 ± 0.4 ( ) ± 0.5 ( ) 18 Heightofbraincase 11.7 ± 0.4( ) ± 0.4( ) 19 Length of diastema 10.8 ± 0.6( ) ± 0.5( )20 Postpalatal length 13.0 ± 0.6 ( ) ± 0.5 ( ) 20 Length of incisive foramina 7.9 ± 0.3 ( ) ± 0.4 ( ) 20 Breadth of incisive foramina 2.8 ± 0.2 ( ) ± 0.2 ( ) 20 Length of bony palate 8.3 ± 0.3 ( ) ± 0.4 ( ) 20 Breadth of palatal bridge at MI 4.1 ± 0.2 ( ) ± 0.2 ( ) 20 Length of bulla 6.7 ± 0.2 ( ) ± 0.2 ( ) 20 Crown length of M ± 0.2 ( ) ± 0.2 ( ) 20 Breadth of Ml 2.0 ± 0.1 ( ) ± 0.1 ( ) 20 a Mean plus or minus one standard deviation, observed range in parentheses, and number of specimens are listed. Males and females are combined in the samples of R. baluensis and R. hoogerwerfi. b Holotype. the throat, chest, and inguinal region. Sides of the head and body are slightly paler than the back and no sharp demarcation exists between coloration of dorsum and venter. The hairs that make up the dorsal pelage are dark gray for most of their lengths and tipped with ochraceous-tawny. The coat is thick, lax, and long; the hairs of the overfur along the back reaching mm in length. Longer and slightly stiffer black guard hairs extend beyond the overfur by mm. The ventral coat is also thick and long (10-12 mm) with the hairs being gray and tipped with white. Fur texture and length is typical of species of Rattus living in cool and wet tropical mountain forest. The ears are small, brown, and densely covered with fine, short brown hairs. Dorsal surfaces of the front and hind feet are brown with a darker brown strip extending the length of each metacarpal and metatarsal surface onto the bases ofthe digits. The claws are short, recurved, and sharp. The convex dorsal surface of each claw is covered with a tuft of long silvery hair. Palmar and plantar surfaces are pale brown and naked. Most of the palmar surface is formed of five large fleshy pads (three interdigital and two metacarpal), each adorned with transverse and semicircular lamellae. The hind foot is long and slender, widening distally to the base of the digits. Four large interdigital pads and two metatarsal pads are fleshy and thick and also inscribed with lamellae. The brown, monocolored tail is much longer than head and body (tables 1 and 2). Circlets of small overlapping scales cover its surface (13-14 rings ofscales per cm, counted about a third of the way from the base of the tail). Three dark brown hairs emerge from

7 1986 MUSSER: SUNDAIC RATTUS 7 the anterior margin of each scale. Near the base of the tail the hairs are twice as long as a scale but distally the scales decrease in size, the tail diameter becomes smaller, the hairs are longer, and the tail more densely covered. Brown and silver hairs extending beyond the tip of the tail to form a brush 3-4 mm long, prompting Robinson and Kloss (1918, p. 53) to write that it was "almost penicillate." Females have one pectoral and two inguinal pairs of teats. Rattus baluensis resembles R. korinchi in size ofbody and hind feet (table 1); coloration of fur over upperparts; length, density, and texture of the coat; color of ears and feet; as well as configuration of palmar and plantar surfaces and number of pads. In contrast to the Sumatran animal, R. baluensis has buffy gray underparts and an absolutely and relatively shorter tail (tables 1 and 2). There are also fewer rings of scales on the tail of R. baluensis (10-12 in 25 adults contrasted with in R. korinchi), and no specimen ofr. baluensis studied has a tail tuft. Female Rattus baluensis have five pairs of teats, one pectoral, one postaxillary, two abdominal, and one inguinal. The cranium of Rattus korinchi is moderately large with a squarish braincase and a slender rostrum (figs. 3, 4; table 1). Structural details of the orbit, alisphenoid region, palatal bridge, mesopterygoid and pterygoid fossae, and pattern of basicranial arterial circulation resemble those of R. rattus, whose features have been described and illustrated by Musser (1981, 1982) and Musser and Newcomb (1983). Aside from these regions, a characteristic ofthe cranium seen from dorsal view is its slender rostrum, which, except for the slightly protruding nasolacrimal capsules, hardly tapers from back to front. Another feature is the narrow interorbital region and the hourglass outline posterior to the orbits that are bounded by ridges extending back along dorsolateral margins of the braincase to the occiput. Prominent behind the orbits, the ridges decrease to an inconspicuous beading along posterior edges of the braincase. From the union ofthe frontals and parietals, these temporal ridges are nearly parallel and give the braincase a rectangular conformation. The sides of the braincase are not vertical but have a gentle medial slope from each squamosal root to temporal ridge. TABLE 2 Ratios (in Percent) of Some Measurements from Samples of Rattus baluensis, Rattus korinchi, and Rattus hoogerwerfia R. R. R. hooger- Ratio baluensis korinchi werfi LLHTB LHB , BR LR ~58 51, LIF LIF 73 78,77 74 LB PLB 52 48,45 44 LB GLB 16 15,15 15 CLMI-3 LBP 81 93,87 89 CLMI-3 GLS 16 19, a Ratios are derived from data listed in table 1. Abbreviations: BR, breadth of rostrum; CLM1-3, crown length of maxillary toothrow; GLS, greatest length of skull; LB, length of bulla; LD, length of diastema; LHB, length of head and body; LIF, length of incisive foramina; LBP, length of bony palate; LR, length of rostrum; LT, length of tail; PPL, postpalatal length. The zygomatic arches are nearly straight and parallel to one another. The top of the cranium above the zygomatic plates and orbit is slightly concave, which gives that portion of the cranium anterior to the braincase a flattened configuration, clearly seen in lateral view (fig. 4). The rostrum is high, each zygomatic plate moderately wide, and the bullae small relative to size of the braincase (tables 1 and 2). The incisors are orthodont and the molar rows are relatively long and terminate anteriorly near the level of the posterior edge of the zygomatic plate. Rattus korinchi's elongate incisive foramina, long maxillary toothrows, and small bullae can be seen in ventral view (fig. 3). Lateral margins of the incisive foramina nearly parallel one another and their posterior margins end less than a millimeter past the anterior alveolar margins of the toothrows. Those molar rows are nearly as long as the bony palate (table 2). The palatal bridge extends

8 8 AMERICAN MUSEUM NOVITATES NO I Fig. 3. Crania of adult Sundaic Rattus compared. Left: R. baluensis (FMNH ), Gunung Kinabalu, Sabah. Middle: R. korinchi (RMNH 23151), Gunung Talakmau, Sumatra. Right: R. hoogerwerfi (USNM ), Gunung Leuser, Sumatra.. All x 1.5. Measurements are listed in table 1. posteriorly beyond the rows of teeth in the form ofa shelf, a characteristic ofmost species of Rattus. In contrast to most Rattus, the auditory bullae are absolutely and relatively smaller (table 2). Although similar in cranial size, Rattus baluensis has a stockier, wider rostrum than that ofr. korinchi (table 2), and the rostrum is not as deep behind the incisors (figs. 3, 4). The nasolacrimal capsules ofr. baluensis are slightly more inflated and each opening is larger. As seen from dorsal view, the ridges bounding the postorbital region sweep back and out resembling the sides of a triangle rather than the bottom half of an hourglass, which is the outline in R. korinchi. The area medial to the dorsal root of each zygomatic arch is swollen in R. baluensis but

9 1 986 MUSSER: SUNDAIC RATTUS 9 much less so in R. korinchi, which accounts for the generally straight but slightly dished dorsal canial profile between braincase and nasal tips in R. korinchi and the slightly convex profile characteristic ofr. baluensis. The way the upper incisors curve slightly back (opisthodont configuration) in R. baluensis contrasts with the orthodont conformation of the teeth in R. korinchi, a difference, like the dorsal cranial outlines, that can be seen in side view. The difference in rostral width between the two species is evident in ventral view. Other contrasts in dimensions and proportions can be appreciated (see also tables 1 and 2). Rattus baluensis has, on the average, shorter incisive foramina that are also shorter relative to length of diastema; much shorter toothrows that occupy less of the bony palate and have anterior margins set well behind posterior edges ofthe zygomatic plates; and larger auditory bullae, not only in actual size but relative to size of braincase. The magnitude of the difference in length of molar row between R. baluensis and R. korinchi is impressive. The contrast in bullar size is also conspicuous and there is narrow overlap in the range of variation between the two samples; only 3 out of the 25 specimens of R. baluensis have a minimum bullar length of 6.4 mm, which is the size of the largest bulla in the two examples of R. korinchi (table 1). The conformation of each dentary of Rattus korinchi is similar to that in other species of Rattus (Musser, 198 1). The bony sheath enclosing the incisor is not elongate, the coronoid process is large and curved back, and the posterior dentary margin is deeply concave (fig. 4). The general shape of each dentary of R. baluensis is similar to that of R. korinchi but is overall slightly smaller, the coronoid process is shorter and less acute, the region anterior to the molar row is slightly deeper, and the toothrow is much shorter. As they are in other species of Rattus, the upper and lower incisors of Rattus korinchi have smooth surfaces. The enamel layers are pale orange. The incisor enamel ofr. baluensis is pigmented more intensely-a deep orange. Many characters in each species, such as the number of roots anchoring each tooth, the degree one tooth overlaps the other, sizes I, 0 I,"Mrlw., domw was molp",.71.. '.. YO..Aww _.mt Fig. 4. Lateral views of crania and mandibles of specimens shown in figure 3. Top: R. baluensis. Middle: R. korinchi. Bottom: R. hoogerwerfi. All x1.5. of molars relative to one another, height of crowns, general cusp patterns, and no posterior cingula on first and second upper molars, are similar to those found in Rattus rattus and many other members of the genus (Musser, 1981, 1982; Musser and Newcomb, 1983). Rattus korinchi and R. baluensis are

10 10 AMERICAN MUSEUM NOVITATES NO TABLE 3 Presence (+) or Absence (-) of Certain Cusps and Cusplets on Molars in Rattus baluensis, Rattus korinchi, and Rattus hoogerwerfia R. R. R. hooger- Trait baluensis korinchi werfi Cusp t3 on M (34) 50 (1) 70 (14) _ 40 (23) 50 (1) 30 (6) Cusp t3 on M3 100 (57) 100 (2) 100 (20) Anterior labial cusplet on M1-100 (56) 100 (2) 100 (20) Posterior labial cusplet on M (54) 100 (2) 95 (19) - 4 (2) 5 (1) Anterolabial cusp on M (56) 100 (2) 94 (17) (1) Posterior labial cusplet on M (55) 50 (1) 83 (15) - 2 (1) 50 (1) 17 (3) Anterolabial cusp on M (56) 100 (2) 78 (14) (4) Anterocentral cusp on M (2) (57) (20) a Number of cusps and cusplets are expressed as percentages; number of specimens is in parentheses. also alike in occlusal topography and frequency of occurrence of the cusps and cusplets that are known to vary in both size and number (figs. 5, 6; table 3). For example, in about half the sample of each species, cusp t3 is present on each second upper molar but cusp t3 is absent from the third upper molar ofall specimens in both samples. Both species lack an anterior labial cusplet but have a posterior labial cusplet on each first lower molar. The anterolabial cusp is present on each second lower molar in all specimens of each species and the posterior labial cusplet ofthat tooth occurs in one of the two specimens of R. korinchi and in all the examples of R. baluensis. The anterolabial cusp is also present on each third lower molar in all specimens of both species and the posterior labial cusplet is part of the third molar in all R. baluensis but not the two specimens of R. korinchi. In addition to the impressive differences in toothrow dimensions between R. korinchi and R. baluensis (table 1), the two species can be distinguished by other dental traits. One complex involves cusps t8 and t9 of the first and second upper molars. On each of these teeth there is a lingual, cusplike ridge (resembling a cusp t7) that extends anterior to touch the back of cusp t4. When cusps are worn to the level seen in RMNH 23151, there is a prominent crescentic ridge connecting the lingual side ofcusp t8 with the posterior margin of cusp t4 on each first and second molar (fig. 5). A comparable lingual ridge developed from cusp t8 does not occur in R. baluensis. Cusp t9, on the labial side of cusp t8, is large and directed more anteriorly in R. korinchi than is cusp t9 on the first and second molars of R. baluensis. Furthermore, cusp t9 of the second upper molar in R. korinchi is much smaller relative to size of the adjacent cusp t8 than is the relatively larger cusp t9 in R. baluensis. There is another occlusal topographic difference between the two species. In R. korinchi, either a mound or cusplet is present on the cingular margin of each first upper molar between cusps t3 and t6. No such structure occurs in R. baluensis. Finally, in both specimens of R. korinchi there is a small but prominent anterocentral cusp on the anterior face of each first lower molar (fig. 6). This cusp is discrete in the holotype, which has moderately worn teeth, but indistinct and represented by a slight bulge in the worn molars ofthe older RMNH An anterocentral cusp is absent in all the specimens of R. baluensis I examined; the front lamina ofeach first lower molar is composed only of large anterolingual and smaller anterolabial cusps. In summary, although superficial inspection suggests that R. baluensis and R. korinchi are alike, careful examination reveals conspicuous morphological differences. Compared with R. korinchi, R. baluensis has a nonpenicillate tail that is shorter relative to length of head and body, a wider rostrum with more inflated nasolacrimal capsules, a

11 1986 MUSSER: SUNDAIC RATTUS I1I different dorsal cranial profile, shorter incisive foramina, absolutely and relatively larger tympanic bullae, opisthodont upper incisors, smaller coronoid processes, absolutely and relatively shorter toothrows, and less cuspidate molars. RELATIONSHIPS TO OTHER RATTUS AND ZOOGEOGRAPHIC SIGNIFICANCE The morphological features ofr. baluensis are similar to those that characterize species of the Rattus rattus group, particularly R. tiomanicus. Rattus tiomanicus is indigenous to the Sunda Shelf and occurs on Peninsular Thailand and Malaysia, Sumatra, Java, Borneo, and many small islands on the Shelfexcept Bali. It has not been recorded from peninsular Burma and Thailand north of the Isthmus of Kra (10 30'N). Off the Shelf, R. tiomanicus lives in the Maratua Archipelago, to the east of Borneo, and Pulau Enggano, southwest of Sumatra (Musser and Califia, 1982). Morphology of skins and skulls indicates that samples of R. mindorensis from Mindoro Island in the Philippines, R. burrus from the Nicobar islands, and R. simalurensis from Pulau Simalulue and nearby islands (table 6) represent insular populations of R. tiomanicus found off of the continental shelf (Musser et al., MS). As might be expected of a species with a discontinuous distribution on a vast number of islands, there is appreciable geographic variation among samples of R. tiomanicus, particularly in body size and pelage coloration. In the Bornean region, for example, the mainland populations consist ofsmall animals with brown upperparts and white venters but the rats living on islands in the Maratua Archipelago, offthe east coast of Borneo and clear of the continental shelf, are large and richly pigmented (Musser and Califia, 1982). Yet wherever R. tiomanicus is found, there are basic pelage features and a cranial conformation that identify the species. These characteristics are also shared by R. baluen- Sis Ṫhe body and tail proportions of R. baluensis (tables 1, 4), number of mammae (five pair), configuration of the cranium (as well as the mandible) (fig. 7), size and projection of incisors, relative size of molars and topography oftheir occlusal patterns (fig. 8) are more like these features in R. tiomanicus than in R. korinchi. The Kinabalu rat is larger and has much longer, thicker, and darker fur than R. tiomanicus from mainland Borneo living at lower elevations, and the thick dark coat separates R. baluensis from most populations of R. tiomanicus on the Sunda Shelf. Body siize as well as cranial and toothrow dimensions, however, overlap broadly between the sample of R. baluensis and some samples of R. tiomanicus; just contrast the values in table 1 with those listed in table 4 for R. tiomanicus from Pulau Maratua. Chasen (1940, p. 158) recognized this morphological link between the Kinabalu animal and R. tiomanicus and wrote that R. baluensis was "almost certainly only a high level form of R. rattus." Chasen was referring to what we now call R. tiomanicus because when he compiled his handlist of Malaysian mammals many of the taxa currently associated with R. tiomanicus (see Musser and Califia, 1982, p. 24) were then regarded as subspecies of R. rattus. Chasen was not thinking of R. rattus diardii, which is the only R. rattus that is found on Borneo, because he recognized it as a commensal houserat in his list. The houserat is not native to Borneo (Medway, 1965) or anyplace on the Sunda Shelf (Musser and Newcomb, 1983). The species is common in lowlands of Sabah wherever there are towns, villages, and gardens (Lim and Heyneman, 1968; Medway, 1965). Specimens ofr. rattus diardii are similar to those ofr. baluensis in body size and some proportions (table 4) but contrast with R. baluensis in the same ways that they differ from examples ofr. tiomanicus (Musser and Califia, 1982, pp ). The shape of the cranium as viewed dorsally is one of these features (fig. 7). From that viewpoint, the supraorbital ridges flow back on either side of the skull in smooth, vase-shaped curves and disappear near the occiput in R. baluensis and R. tiomanicus. In R. rattus diardii, the supraorbital ridges are high and wide and form a conspicuous angular shelfbehind the interorbital area and then sweep back over the sides of the braincase to the occiput, forming, especially in old animals, a square or rectangular configuration.

12 12 AMERICAN MUSEUM NOVITATES NO ,..1! "I 4L It.1". c I.../ i,_.p.i., A Fig. 5. Upper molar rows of same specimens ofrattus shown in figure 3. Left: R. baluensis. Middle: R. korinchi. Right: R. hoogerwerfi. Approximately x 10. In middle figure, each arrow on left points to cusp t9, each arrow on right points to crescentic ridge formed by lingual extension of cusp t8 and back of cusp t4. The character information based upon museum specimens suggests that R. baluensis, living in an island of mountain forest on the high slopes of Gunung Kinabalu, is a largebodied, dark, densely furred morphological counterpart of R. tiomanicus, which occurs at middle altitudes on Gunung Kinabalu and in the lowlands of Sabah and regions farther south in Borneo (Medway, 1965; Musser and Califia, 1982). The long, dense fur is certainly an adaptation to low, ambient temperatures; body size may be a similar adaptation, and it may even reflect the association between island area (in this case, the extent of suitable montane habitat) and body size (Heaney, 1978) since mountain forest on Gunung Kinabalu is essentially an island in a vast expanse of lowland tropical evergreen forest -at least where that forest has not been removed. There is no evidence suggesting gene exchange between R. baluensis and the smaller-bodied, brighter R. tiomanicus sabae; rather their similarities indicate that they may have shared a common ancestor. If Rattus baluensis is more closely related to R. tiomanicus than to R. korinchi, then where does that Sumatran species fit in the pattern of phylogenetic relationships among species ofrattus? Rattus korinchi and R. baluensis share many features, as already discussed, that are also shared in common with species ofrattus. They also share long, thick, and dark pelage, but that is probably a convergent adaptation to an ambient environment associated with cool and wet mountain forests in the tropics. Excluding these aspects of its morphology, R. korinchi differs in the same ways from R. baluensis and from R. tiomanicus, suggesting that the Sumatran rat is not part of the same species group that embraces R. baluensis and R. tiomanicus. Rattus korinchi has been linked to another

13 1 986 MUSSER: SUNDAIC RATTUS 13 * 9h Fig. 6. Lower molar rows of same specimens ofrattus shown in figure 3. Left: R. baluensis. Middle: R. korinchi. Right: R. hoogerwerfi. Approximately x 10. Arrow at top of middle toothrow points to outline of anterocentral cusp. Sundaic endemic, R. hoogerwerfi, which was described by Chasen in The species was subsequently characterized in greater detail by Miller (1942), who reported upon a much larger sample. All specimens (ANSP and ; MZB , 4854-holotype, and ; RMNH ; USNM ) come from the Aceh region of northern Sumatra; most were trapped between 2900 and 9300 ft on the forested foothills and upper slopes of Gunung Leuser. Chasen (1939, p. 208) provided this opinion about the affinities of R. korinchi: "perhaps the nearest relative of R. hoogerwerfi is Rattus baluensis korinchi. If consideration of the tail is excluded, the two forms have some superficial resemblance to each other, and the peculiar characters ofthe skull ofhoogerwerfi are to some extent adumbrated in adults of korinchi." Chasen excluded the tail because in R. hoogerwerfi it is white for about the distal half, dark brown over the basal half, and not penicillate; the slightly brushy tail of R. korinchi is monocolored dark brown; in both species the tail is very long relative to length of head and body (tables 1, 2). Features ofthe tail are not the only external contrasts between the two species. Rattus hoogerwerfi is larger in body size than R. korinchi (table 1), has eight teats, and has ochraceous brown upperparts and bright buffy to ochraceous gray underparts; the overall coloration of the fur is much richer than the pelage of R. korinchi. Morphology and coloration of the ears and feet are similar in the two species, as is density and texture of the fur İn his diagnosis of R. hoogerwerfi, Chasen (1939, p. 207) noted that the "nasals and the interorbital region of the skull is unusually flattened." This is one of the diagnostic cranial characters that Chasen also thought was typical ofr. korinchi. However, I cannot see the resemblance; the area above the dorsal root of each zygomatic plate is swollen in R.

14 14 AMERICAN MUSEUM NOVITATES NO Fig. 7. Crania of adult Rattus from the Bornean region. From left to right: R. rattus diardii (USNM ), Gunung Kinabalu, 4900 ft, Sabah; R. baluensis (FMNH ), Gunung Kinabalu, 11,200 ft, Sabah; R. tiomanicus sabae (USNM ), 3300 ft, Sabah; R. tiomanicus mara (USNM ), Pulau Maratua. All x 1.5. hoogerwerfi to the same extent as is seen in specimens of R. baluensis (fig. 4). Furthermore, the frontal bones between the orbits are also swollen and unlike the flattened, slightly concave dorsal profile characteristic of R. korinchi (fig. 4). The two species do share cranial and dental characters but some of these are common to nearly all species ofrattus (Musser and Newcomb, 1983) and provide no information about the relationships of R. korinchi and R. hoogerwerfi. Both have, for example, similar configurations of the orbit, alisphenoid region, palatal bridge, as well as the same pattern of basicranial arterial circulation. Both species share the same number of roots an-

15 1986 MUSSER: SUNDAIC RATTUS 15 Fig. 8. Upper (two on left) and lower (two on right) molar rows of Rattus. Left view of each couplet is Rattus baluensis from Gunung Kinabalu (FMNH ; CLM1-3 = 6.8 mm, CLM-3 = 6.6 mm), right view in each couplet is Rattus tiomanicus from Java (AMNH ; CLM'-3 = 7.0 mm, CLMI3 = 7.2 mm). Note similarity in cusp patterns. A primary but minor difference is the frequency of occurrence of cusp t3 (arrow) on first and second upper molars. Cusp t3 is present on each first molar in 60 percent of the sample of R. baluensis and absent from each second molar of all specimens (table 3). In 20 specimens of R. tiomanicus from Borneo, cusp t3 occurs on each first molar in 18 (90%) rats and on each second tooth in 9 (45%) specimens. These frequencies are typical for the Javan sample as well as most other samples of R. tiomanicus. choring each molar, the degree one tooth overlaps the other, sizes of molars relative to one another, height of crowns, general cusp patterns; both lack posterior cingula on first and second upper molars. Rattus hoogerwerfi and R. korinchi also possess narrow rostra, as well as small bullae and long molar rows relative to size of cranium (tables 1, 2), and orthodont configuration ofupper molars (fig. 4). Small auditory bullae are thought to be primitive in murines (Musser and Newcomb, 1983) and do not necessarily reflect a close phylogenetic relationship between the two species. Conformation of the rostrum and incisors and relative molar dimensions may indicate a closer affinity between R. hoogerwerji and R. korinchi than between either of those two species and R. baluensis. However, the polarities of these characters are unclear and may not affirm a close link between the two Sumatran species. The conspicuous differences between R. hoogerwerfi and R. korinchi are more notable than their few similarities. In addition to the coloration (bicolored versus monocolored) and pilosity (nonpenicillate versus slightly penicillate) of the tail, and the dorsal outline of the cranium (slightly convex from nasal tips to interorbit in R. hoogerwerfi but flat to slightly concave in R. korinchi; fig. 4), the species contrast in the following features. 1. The nasal bones ofr. hoogerwerfi have

16 16 AMERICAN MUSEUM NOVITATES NO TABLE 4 Measurements (in Millimeters) and Ratios (in Percent) of Adult Rattus rattus diardii and Rattus tiomanicus from Borneo and the Maratua Archipelagoa R. tiomanicus Meas. and Ratio R. rattus, Sabah Sabah South Kalimantan Pulau Maratua LHB ± 16.6 (15) ± 14.6 (5) ± 12.2 (16) ± 31.5 (30) LT ± 23.8 (15) (5) ± 10.2 (16) ± 16.7 (30) GLS 42.1 ± 1.9 (15) 37.7 ± 1.9 (5) 38.1 ± 1.4 (16) 43.1 ± 2.1 (30) LPB 8.6 ± 0.6 (15) 7.8 ± 0.4 (5) (16) 9.0 ± 0.5 (30) PPL 14.9 ± 1.0 (15) 12.9 ± 1.0 (5) 12.4 ± 0.7 (16) 14.5 ± 0.9 (30) LB 7.0 ± 0.4 (15) 6.4 ± 0.2 (5) 6.8 ± 0.3 (16) 6.9 ± 0.2 (30) CLM (15) 6.2 ± 0.2 (5) 6.4 ± 0.2 (16) 6.6 ± 0.2 (30) LT LHB LB GLS CLM1-3 LPB CLM GLS a Mean plus or minus one standard deviation, number of specimens in parentheses, and observed range are listed. Abbreviations of measurements are explained in table 2. Fig. 9. Nasolacrimal capsules in Rattus hoogerwerfi (left) and Rattus korinchi (right). Located just anterior to the zygomatic plate, the capsule of R. hoogerwerfi has a nearly horizontal opening, while that of R. korinchi is aligned at about 45. Drawings from same specimens shown in figure 4.

17 1 986 MUSSER: SUNDAIC RATTUS 17 Fig. 10. Pterygoid fossae in Rattus korinchi (left) and Rattus hoogerwerfi (right). Excavation of the fossa in R. korinchi begins at the lateral margin, but the medial half of the fossa is deeply excavated and the lateral half is nearly flat in R. hoogerwerfi. Note the different conformation of each pterygoid bridge (ptb), posterior opening of each alisphenoid canal (pa), extent of foramen ovale (fo) exposed, and position of each transverse canal (tc) relative to the bony eustachian tube (et) of the auditory bulla (ab). See text for an expanded discussion of pterygoid contrasts between the species. Spt = sphenopterygoid vacuity; other structures and foramina not labeled can be identified by referring to appropriate illustrations in Musser and Newcomb (1983). Drawn from same specimens shown in figure 3. an oblanceolate outline whereas those of R. korinchi are linear from dorsal perspective (fig. 3). 2. The rostrum just posterior to the incisors is not as deep in R. hoogerwerfi as it is in R. korinchi (fig. 4). 3. The nasolacrimal capsules are slightly wider in R. hoogerwerfi and each one opens dorsad. Each opening is aligned at about a 450 angle in R. korinchi (fig. 9). 4. The braincase ofr. hoogerwerfi is wider than in R. korinchi (table 1) and oval in dorsal outline as contrasted with a more boxlike conformation in R. korinchi (fig. 3). 5. The configuration of the pterygoid fossae, which is the site of origin of the internal pterygoid muscles, differs (fig. 10). In R. hoogerwerfi, each fossa is only slightly higher than the bony palate and the lateral half of the pterygoid shelf is nearly flat for about three-fourths of the distance from the posterior margin of the palate to the auditory bulla. Anterior to the bulla and at the place where the shelf merges with the pterygoid bridge, the surface turns sharply upward and in that slope are two openings. The lateral one is the posterior opening of the alisphenoid canal, which is vertical, almost like the entrance to a tube. Medial of that but lateral to the bony eustachian tube, is the transverse canal. Anterior to these openings, the pter-

18 18 AMERICAN MUSEUM NOVITATES NO ygoid plate is perforated by a wide sphenopterygoid vacuity. At the level ofthe posterior opening ofthe alisphenoid, the lateral margin ofeach pterygoid plate expands to form a low and broad mound, the pterygoid bridge. In R. korinchi, the pterygoid fossae are excavated beginning at their lateral margins. Each surface is breached by a wide sphenopterygoid vacuity and the lateral half is not nearly flat but slopes upward toward the midline of the cranium. About halfway between margin of the bony palate and tympanic bullae, the pterygoid surface gradually slopes upward. Here there are two openings. The medial one is the entrance to the transverse canal, which is just above the bony eustachian tube (located lateral to the tube in R. hoogerwerfi). The spacious, nearly horizontal lateral opening is the posterior opening ofthe alisphenoid canal. Because the opening is nearly horizontal, from a ventral perspective (fig. 10) you can see into the posterior portion of the foramen ovale, which is located in the side of the braincase above the pterygoid fossa. The edge of the pterygoid surface lateral to the opening into the alisphenoid canal-the pterygoid bridge-is formed by a high and narrow ridge. The conformation of the pterygoid plates and openings in R. hoogerwerfi has been interpreted as primitive in muroid rodents (Carleton, 1980; Musser and Newcomb, 1983). A more fully excavated fossa, nearly horizontal posterior opening to the alisphenoid canal through which the foramen ovale can be seen, and a high ridge forming the pterygoid bridge, as found in R. korinchi and most Rattus including R. baluensis, are specializations (fig. 10; see also Musser, 1981, 1982; Musser and Newcomb, 1983). 6. The shape of the dentary is similar in R. hoogerwerfi and R. korinchi but the coronoid process is not as large relative to the body ofthe ramus, the bony sheath enclosing the incisor is not as wide or deep, and in any pair of specimens of approximately the same age, the absolute size ofthe dentary is smaller in R. hoogerwerfi, even though the cranium is larger (fig. 4). 7. Upper incisors of R. hoogerwerfi are more deeply pigmented and lowers are smaller in diameter and more gracile than those of R. korinchi. 8. Compared to most Rattus, Rattus hoogerwerfi and R. korinchi have large molars and long molar rows, both in absolute dimensions and relative to lengths of bony palate and cranium (tables 1, 2). The two species also share similar cusp patterns ofthe upper and lower molars (fig. 5; table 3), yet there are differences. The lingual extension of cusp t8 of each first and second upper molar in R. hoogerwerfi extends laterally from the cusp and does not merge with the posterior margin of cusp t4; in R. korinchi, the lingual process of cusp t8 extends anteriorly and contacts the back of cusp t4 to form a crescentic ridge after some wear. Cusp t9, on the labial surface of the large cusp t8, is also directed laterally in R. hoogerwerfi; in R. korinchi the comparable cusp extends anteriorly in both first and second upper molars. There are slight differences in occurrence of certain cusps and cusplets on the lower molars (fig. 6; table 3), but the major contrast is the absence of an anterocentral cusp at the anterior end ofeach first molar in R. hoogerwerfi; both examples of R. korinchi have such a cusp. There is no derived character or set ofcharacters suggesting R. hoogerwerfi to be a sister species of R. korinchi. Their external similarities (long tail relative to head and body, thick and long fur, dark coloration) may be independent adaptations to a scansorial or arboreal life in cool and wet mountain forests. The cranial characters shared by both are either primitive retentions (small bullae relative to cranial dimensions, for example), derived features found in many other species of Rattus, or ambiguous in indicating relationships (narrow rostrum, orthodont upper incisors, large molars relative to size of cranium). The combination of bicolored tail, nearly horizontal alignment ofthe rim ofeach nasolacrimal capsule, and large oval braincase-all derived characters compared with the nature of these structures in most species of Rattus (R. korinchi and R. baluensis, for example) -and conformation of each pterygoid fossa and associated foramina-which is primitive with respect to other species of Rattus- isolate R. hoogerwerfi within the genus. The uniqueness of R. hoogerwerfi was appreciated by Sody, who in 1941 (p. 260) proposed that the species be placed in its own

19 1986 MUSSER: SUNDAIC RATTUS 19 genus, Cironomys, at least partly because of the number and distribution ofmammae (one pectoral pair, one abdominal pair, and two inguinal pairs). To Miller (1942, p. 150), the "Atjehan mountain rat is a member of the peculiar group (nearly related to Rattus rattus and its common Malayan representatives) that includes the Celebean Rattus xanthurus (Gray), R. celebensis (Gray), R. marmosurus (Thomas), and R. hamatus (Miller and Hollister), as well, apparently, as the Philippine R. everetti (Gunther) and the R. macleari (Thomas) of Christmas Island." All these species share medium to large body size, a long bicolored tail, relatively large molars, and distributions on islands; otherwise, there is no set of characters tying R. hoogerwerfi any closer to these species than to any other set of species either within or outside of the morphological limits ofrattus (Musser, MS). The morphological isolation ofr. hoogerwerfi partly illuminates the question about affinities of R. korinchi. The two are not closely related, judged from analyses ofcharacters in the stuffed skin, skull, and dentition. Compared with R. hoogerwerfi, R. korinchi has tail coloration, nasolacrimal capsules, and pterygoid fossae like most species of Rattus. However, judged by the morphological differences between R. korinchi and R. baluensis, the Sumatran rat is not closely allied to species in the R. rattus group, and is morphologically isolated compared with most species in the genus. On Borneo, the morphological attributes of R. baluensis suggest the hypothesis that it and R. tiomanicus shared the same, recent common ancestor. On Sumatra, neither R. korinchi nor R. hoogerwerfi appears to be derivatives of that ancestral population. THE SUNDAIC RATTUS The separation of R. korinchi from R. baluensis brings to 41 the number of murid species now known to be native to the Sunda Shelf (table 5) and slightly changes the percentage of endemic murids on Borneo and Sumatra (compare table 5 with table 36 in Musser and Newcomb, 1983, p. 532). There are five, not four, species of Rattus recognized from the peninsula and islands on the Shelf. Present critical comparisons among specimens of R. baluensis, R. korinchi, R. hoogerwerfi, and R. tiomanicus modify parts of Musser and Newcomb's (1983, pp ) assertions regarding relationships of these native Sundaic species but support other aspects of their observations. On Borneo there are 19 species of native murids, of which two (1 %) are Rattus. One ofthese, R. baluensis, is known to occur only in montane forest on the high slopes of Gunung Kinabalu and in its morphology appears to be closely related to that of R. tiomanicus, which is found at lower elevations on Kinabalu and elsewhere throughout Borneo and on nearby islands. Rattus tiomanicus is also widespread elsewhere on the Sunda Shelf and has been recorded from beyond the 180-m bathymetric line. Through R. tiomanicus, the Kinabalu species appears to be more closely related to R. rattus, R. argentiventer, R. nitidus, and other species in the R. rattus group (which occur primarily in India and Indochina) than to any other species of Rattus endemic to the Sunda Shelf. Ofthe 20 murid species native to Sumatra, four (20%) are Rattus (table 5); two of these also occur elsewhere and two are endemic. Rattus annandalei has been recorded from the mainland of eastern Sumatra, the islands of Padang and Rupat off the coast of eastern Sumatra, peninsular Malaysia, and Singapore (Musser and Newcomb, 1983). Its range, however, is probably more widespread than is indicated by specimens in museum collections. On peninsular Malaysia, it has usually been regarded as uncommon and local in its distribution (Chasen, 1940; Medway, 1969) but Muul and Lim (1971, p. 435) found it common "in lowland secondary forests... and in kampong rubber estates... whenever traps were set low on trees." They also noted that some specimens were caught on the ground and that none were encountered in primary forest. There are no specimen records of R. annandalei from highlands. Rattus tiomanicus is the other native nonendemic species of the genus occurring on Sumatra. It, like R. annandalei, is primarily lowland in distribution but has been recorded from as high as 3600 ft on the slopes of Gunung Leuser (specimens collected there were originally described as a new species, R. blangorum, by Miller, 1942).

20 20 AMERICAN MUSEUM NOVITATES NO TABLE 5 Living Native Murids on Peninsula and Islands of the Sunda Shelfa Peninsular Thailand and Smallerb Palawan Genera and Species Malaya Sumatra Borneo Java Bali Islands Group Rattus tiomanicus I I I I I I I Rattus baluensis - - E Rattus korinchi - E Rattus hoogerwerfi - E Rattus annandalei I I Palawanomysfurvus E Kadarsanomys sodyi E Sundamys muelleri I I I - - I I Sundamys infraluteus - I I Sundamys maxi E Berylmys bowersii 0 0 Mus crociduroides - E Mus vulcani E - Leopoldamys sabanus Leopoldamys edwardsi Niviventer bukit Niviventer cremoriventer I I I I I I - Niviventer rapit I I I Niviventer lepturus E - - Maxomys surifer Maxomys rajah I I I - - I - Maxomys panglima E Maxomys whiteheadi I I I - - I - Maxomys bartelsii E Maxomys inflatus - E Maxomys hylomyoides - E Maxomys inas E Maxomys alticola - - E Maxomys ochraceiventer - - E Maxomys baeodon - - E Chiropodomys gliroides Chiropodomys muroides - - E Chiropodomys major - - E Chiropodomys calamianus E Haeromys margaretae - - E Haeromys pusillus - - E Haeromys sp E Pithecheir melanurus E Pithecheir parvus E - Lenothrix canus I I I Hapalomys longicaudatus 0 Total species Total indigenous 8 (47)C 9 (45) 8 (42) 2 (17) 2 (50) 5 (63) 2 (33) Total also found elsewhere 7 (41) 6 (30) 3 (16) 4 (33) 2 (50) 3 (37) - Total endemic 2 (12) 5 (25) 8 (42) 6 (50) (67) Abbreviations: -, apparently absent; E, endemic to a particular island; I, indigenous to the Sunda Shelf; 0, also found in regions outside the Sunda Shelf. a Defined here, the northern limit of the Sunda Shelf is near the Isthmus of Kra between latitudes 10 and 12 N.

21 1 986 MUSSER: SUNDAIC RATTUS 21 The two Sumatran endemics are R. hoogerwerfi and R. korinchi. Rattus baluensis inhabits a similar ecological situation, living in mountain forests on Gunung Kinabalu in northern Borneo. No such ecologically similar species ofrattus is known from any other island on the Sunda Shelf or on peninsular Malaysia. The only native Rattus on Java and Bali, where there is dramatic topographic relief, is R. tiomanicus. That species and R. annandalei represent the genus on peninsular Malaysia and Singapore. The endemic mountain forest species of Rattus on Sumatra likely had a different evolutionary history than did their ecological counterpart on Gunung Kinabalu. On Borneo, the endemic and montane R. baluensis is apparently related to the widespread and lowland R. tiomanicus. Morphology of the Sumatran R. hoogerwerfi and R. korinchi does not tie either of these species closely to R. tiomanicus and each may represent a separate branching from ancestral Rattus at an earlier date than the later divergence of R. tiomanicus and R. baluensis. If R. hoogerwerfi and R. korinchi are not close to R. tiomanicus and R. baluensis in morphology of characters associated with skins, skulls, and teeth, they are not any more closely related to each other, or to R. annandalei, the other Sundaic Rattus. That species differs from all other Rattus in its retention of many primitive external, cranial, and dental characters combined with highly specialized, large and inflated tympanic bullae. Rattus annandalei, like R. hoogerwerfi and R. korinchi, seems isolated within the present morphological limits of the genus (Musser and Newcomb, 1983). A phylogenetic study of Sundaic Rattus is beyond the scope of the present report. My intention has been to define R. baluensis and R. korinchi by contrasting their morphology and testing previously suggested relationships of each. Two hypotheses are suggested by this research. First, R. baluensis is a sister species of R. tiomanicus. Both share many morphological characteristics with R. rattus and related species occurring in Indochina and the Indian Region (Musser and Newcomb, 1983). Just how closely all these species are related to one another is unknown at present and any resolution to their phylogenetic alliances awaits critical study, especially of geographic variation within R. tiomanicus and an assessment of its affinities to Indochinese R. rattus (Musser and Califia, 1982). The second hypothesis is that R. korinchi is not closely related to either R. baluensis or R. hoogerwerfi as was asserted by Chasen (1940) and Musser and Newcomb (1983). Rattus korinchi does share more derived features with R. tiomanicus and R. baluensis than with R. hoogerwerfi, but the Kerinci rat is not part of the R. tiomanicus-group radiation. It, like R. hoogerwerfi, seems isolated within the morphological framework ofrattus Ṗerhaps the relatives ofr. korinchi and R. hoogerwerfi, as well as R. annandalei, should be looked for in places off of the Sunda Shelf beyond the 180-m bathymetric line. On islands and small archipelagos to the east, south, and west of the Shelf are native Rattus that represent either (1) insular populations of R. tiomanicus, (2) species that are similar to R. tiomanicus in their morphology, or (3) other species characterized by morphologies very different from that typical of R. tiomanicus and its relatives (table 6). Most of the islands have only one representative of the three groups. Pulau Enggano is an exception because one member ofeach group occurs there: R. tiomanicus vernalus;.r. adustus, a relative of R. tiomanicus; and R. enganus, which is not closely allied to the R. tiomanicus group. Species listed in the third group in table 6 share the attributes of insular distributions South of the Isthmus, the 100-fathom line marks the boundary, as indicated by the unbroken red line on the map at the front of Chasen's (1940) handlist of Malaysian mammals. Regions outside the Sunda Shelf include Indochina north of the Isthmus of Kra (10'30'N), the Maratua Archipelago east of Bomeo, the Mentawai Islands, and Pulau Enggano. b Includes the Raiu and Lingga archipelagos, the Natuna Islands, Kepulauan Anambas, Pulau Banka, Pulau Billiton, and Pulau Nias. c Percent in parentheses.

22 22 AMERICAN MUSEUM NOVITATES NO TABLE 6 Distribution of Native Rattus on Islands Located East, South, and West of the Sunda Shelf Beyond the 180-m Bathymetric Linea Taxon Populations of R. tiomanicus R. mindorensis R. tiomanicus mara R. simalurensis R. burrus R. tiomanicus vernalus Species closed allied with R. tiomanicus R. adustus R. lugens?r. palmarum Species not closely related to R. tiomanicus R. everetti R. tawitawiensis R. hoffmanni R. xanthurus R. macleari R. nativitatis R. enganus R. stoicus Place Mindoro Island, the Philippines Maratua Archipelago, east of Bomeo Islands of Babi, Lasia, Siumat, and Simalule Nicobar islands, northwest of Sumatra Pulau Enggano, southwest of Sumatra Pulau Enggano Mentawai islands, southwest of Sumatra Nicobar islands Philippine islands, from Luzon to Mindanao Tawitawi Island, Sulu Archipelago Sulawesi Sulawesi and Pulau Peleng Christmas Island, south of Java Christmas Island Pulau Enggano Andaman islands, south of Burma a Distributions, along with maps, and morphological characteristics of the species are discussed by Musser and Califia (1982), Musser and Newcomb (1983), and Musser and Heaney (1985). combined with unique morphologies making them unlike most other species ofrattus occurring on islands and peninsulas of the Sunda Shelf and on continental Asia and India. All eight, like R. hoogerwerfi, R. korinchi, and R. annandalei, are morphologically and probably phylogenetically peripheral to the main cluster of species in Rattus (Musser et al., MS). Some of the species are terrestrial, others are arboreal; all are forest dwellers. Each one is defined by a unique set of characteristics and each retains a greater number of primitive features compared with Rattus rattus, R. tiomanicus, and allied species. The distributions and morphological characteristics suggest that the species represent island relicts ofan ancestral radiation that took place earlier than the one producing R. tiomanicus and its relatives. Future critical study of native Sundaic Rattus should also include samples from the eight species in table 6. Such a study should also include Rattus montanus from the highlands of Sri Lanka (Phillips, 1981). McKay (1984, p. 422) has speculated that R. montanus is "possibly a sample of extremely large R. rattus kelaarti males," but the holotype ofr. montanus and other specimens I have studied clearly represent a species with very different morphology than is usual for R. rattus. In body size and fur color, R. montanus closely resembles R. korinchi. The two species also share similar conformations of cranium and mandible and both have long toothrows. Rattus montanus has more cuspidate molars with a discrete posterior cingulum at the back of each first upper molar; overall, the molar patterns ofr. montanus are more primitive than that seen in R. korinchi. The perception that R. baluensis and R. korinchi are separate species alters our view ofthe native Sundaic murid fauna. Now, five species are known only from Sumatra: Rattus korinchi, Rattus hoogerwerfi, Mus crociduroides, Maxomys inflatus, and Maxomys hylomyoides. Eight species are endemic to Borneo: Rattus baluensis, Maxomys alticola, Maxomys ochraceiventer, Maxomys baeodon, Chiropodomys muroides, Chiropodomys major, Haeromys margaretae, and Haeromys pusillus. Before baluensis was separated from ko-

23 1 986 MUSSER: SUNDAIC RATTUS 23 rinchi, there were two species common to Borneo and Sumatra only. Now there is but one: Sundamys infraluteus, a highland rat that has been collected in the upper part of lowland tropical evergreen forest and higher in montane forest. It has been recorded from five places along the mountain backbone of Sumatra, from two mountains in Sabah, and from one in Sarawak (Anderson et al., 1982; Musser and Newcomb, 1983). Although there are slight differences in coloration and size between samples from the two islands, the samples have the aspect ofa species that once consisted of interbreeding populations. Musser and Newcomb (1983) considered them insular populations of one species. Although found mostly in the mountains, specimens from both Sumatra and Borneo come from as low as 2300 and 3000 ft. When the Sunda Shelfwas entirely above sea level and drained by great rivers during the Pleistocene, temperatures were cooler, forest formations occurred at lower elevations (Whitten et al., 1984), and populations of S. infraluteus were probably continuous in some regions. The morphological differences contrasting the Bornean and Sumatran populations of Sundamys infraluteus do not resemble either the magnitude or kinds of differences that distinguish Rattus baluensis and R. korinchi. LITERATURE CITED Allen, G. M., and H. J. Coolidge, Jr Mammals. In Mammal and bird collections of the Asiatic Primate Expedition. Bull. Mus. Comp. Zool. Harvard College, vol. 87, no. 3, pp , 3 maps. Anderson, J. A. R., A. C. Jermy, and the Earl of Cranbrook Gunung Mulu National Park: a management and development plan. London, Royal Geog. Soc., pp. i-viii, 1-345, figs. 1-6, maps 1-8. Carleton, Michael Dean Phylogenetic relationships in Neotomine-Peromyscine rodents (Muroidea) and a reappraisal of the dichotomy within New World Cricetinae. Misc. Publ. Mus. Zool., Univ. Mich., no. 157, pp. i-vii, 1-146, figs Chasen, F. N Two new mammals from North Sumatra. Treubia, vol. 17, pt. 3, pp A handlist of Malaysian mammals (a systematic list of the mammals of the Malay Peninsula, Sumatra, Borneo and Java, including the adjacent small islands). Bull. Raffles Mus., Singapore, Straits Settlements, no. 15, pp. i-xx, 1-209, 1 map. Corner, E. J. H Royal Society Expedition to North Borneo 1961: Rpt. Linnean Soc. London, vol. 175, pt. 1, pp. 9-56, figs. 1-4, pls Ellerman, J. R The families and genera of living rodents. London, British Museum (Nat. Hist.), vol. 2, Family Muridae, pp. i- xii, 1-690, figs The families and genera of living rodents. London, British Museum (Nat. Hist.), vol. 3, pt. 1, pp. i-v, Heaney, Lawrence R Island area and body size of insular mammals: evidence from the tri-colored squirrel (Callosciurus prevosti) of Southeast Asia. Evolution, vol. 32, pp , figs Jacobs, M Plant life on Mount Kinabalu. Malayan Nature Jour., vol. 15, pp , pl. 31, 1 map. Lim, Boo-Liat, and D. Heyneman A collection of small mammals from Tuaran and the southwest face of Mt. Kinabalu, Sabah. Sarawak Mus. Jour., vol. 16, nos , pp , pls McKay, G. M Ecology and biogeography ofmammals. In C. H. Fernando (ed.), ecology and biogeography in Sri Lanka. The Hague, Martinus Nijhoff Publishers, pp , 1 fig. Medway, Lord Mammals of Borneo. Field keys and an annotated checklist. Singapore, Malay. Branch Roy. Asiatic Soc., Malaysia Printers, pp. i-xiv, 1-193, figs. 1-9, pls. 1-34, 1 map The wild mammals of Malaya and offshore islands including Singapore. Kuala Lumpur and Singapore, Oxford Univ. Press, xix pp., figs. 1-1, pls Mammals of Borneo. Field keys and an annotated checklist. Monographs of the Malaysian Branch of the Royal Asiatic Society, no. 7, Kuala Lumpur, Perchetakan Mas Sdn. Bhd., xii + 172, pp. figs pls

24 24 AMERICAN MUSEUM NOVITATES NO Miller, Gerrit S., Jr Catalogue of the mammals of Western Europe (Europe exclusive of Russia) in the collection of the British Museum. London, British Museum (Nat. Hist.), pp. i-xv, , figs Zoological results of the George Vanderbilt Sumatran Expedition, Part V. Mammals collected by Frederick A. Ulmer, Jr. on Sumatra and Nias. Proc. Acad. Nat. Sci. Philadelphia, vol. 94, pp , pl Misonne, Xavier African and Indo-Australian Muridae. Evolutionary trends. Mus. Roy. l'afrique Cent., Tervuren, Zool., no. 172, pp , figs. A-K, pls Musser, Guy G Results of the Archbold Expeditions. No Notes on systematics of Indo- Malayan murid rodents, and descriptions of new genera and species from Ceylon, Sulawesi, and the Philippines. Bull. Amer. Mus. Nat. Hist., vol. 168, art. 3, pp , figs Results of the Archbold Expeditions. No The definition of Apomys, a native rat of the Philippine Islands. Amer. Mus. Novitates, no. 2746, pp. 1-43, figs Musser, Guy G., and Debra Califia Results of the Archbold Expeditions. No Identities of rats from Pulau Maratua and other islands off east Borneo. Amer. Mus. Novitates, no. 2726, pp. 1-30, figs Musser, Guy G., and Lawrence R. Heaney Philippine Rattus: a new species from the Sulu Archipelago. Amer. Mus. Novitates, no. 2818, pp. 1-32, figs Musser, Guy G., and Cameron Newcomb Malaysian murids and the giant rat of Sumatra. Bull. Amer. Mus. Nat. Hist., vol. 174, art. 4, pp , figs Muul, Illar, and Lim Boo-Liat New locality records for some mammals ofwest Malaysia. Jour. Mammal., vol. 52, no. 2, pp , 1 fig. Phillips, W. W. A Manual of the mammals of Sri Lanka. Part II. Wildlife and Nature Protection Society of Sri Lanka, pp. i-xl, , figs , pls Robinson, Herbert C., and C. Boden Kloss Preliminary diagnoses of some new species and subspecies ofmammals and birds obtained in Korinchi, West Sumatra, Feb.-June Jour. Straits Branch Roy. Asiatic Soc., no. 73, pp Results of an Expedition to Korinchi Peak, 12,400 ft., Sumatra. Mammals of Korinchi. Jour. Fed. Malay States Mus., vol. 8, pt. 2, pp. 1-72, 1 pl On mammals chiefly from the Ophir District, West Sumatra. Ibid., vol. 7, pt. 4, pp Mammals of Korinchi and of Sumatra: Addenda and Corrigenda. Ibid., vol. 8, pt. 2, pp , pls Sody, H. J. V On a collection of rats from the Indo- Malayan and Indo-Australian regions (with descriptions of 43 new genera, species and subspecies). Treubia, vol. 18, pt. 2, pp Tjia, H. D The Sunda Shelf, Southeast Asia. Z. Geomorph. N. F., vol. 24, no. 4, pp , figs Thomas, Oldfield On the mammals of Mount Kina Balu, North Borneo. Z. Geomorph. N. F., pp , 1 pl A preliminary revision of the Bornean species of the genus Mus. Ann. Mag. Nat. Hist., vol. 14, ser. 6, pp Whitmore, T. C Tropical rain forests ofthe Far East, 2nd ed. Oxford, Oxford University Press, xvi pp., numerous figs. Whitten, Anthony J., Sengli J. Damanik, Jazanul Anwar, and Nazaruddin Hisyam The ecology of Sumatra. Gadjah Mada University Press, xiv pp., numerous figs. and pls. L Recent issues of the Novitates may be purchased from the Museum. Lists of back issues of the Novitates, Bulletin, and Anthropological Papers published during the last five years are available free of charge. Address orders to: American Museum of Natural History Library, Department D, Central Park West at 79th St., New York, New York i